Close cooperation with Lorenzo Bianconi resulted
in these patches which fix all remaining seen issues
when using dynack.
Fix link losses when:
- Late Ack's are not seen or not present
- switching from too low static coverage class to dynack on a live link
These are fixed by setting the Ack Timeout/Slottime to
the max possible value for the currently used channel width when
a new station has been discovered.
When traffic flows, dynack is able to adjust to optimal values
within a few packets received (typically < 1 second)
These changes have been thoroughly tested on ~60 offshore devices
all interconnected using mesh over IBSS and dynack enabled on all.
Distances between devices varied from <100m up to ~35km
[move patches to correct folder + renumber]
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
(cherry picked from commit f6e8ba0238fe349b7529357793e2fb18635819ed)
This fixes:
1) Crash during USB disconnect
2) Crash in brcmf_txfinalize() on rmmod with packets queued
3) Some errors in exit path
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
1) Crash/Oops fixes
2) One-line patch for BCM43456 support
3) Fix communication with some specific FullMAC firmwares
4) Potential fix for "Invalid packet id" errors
5) Important helper for reporting FullMAC firmware crashes
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
This partitialy reverts commit f506de2cda.
Registering the GPIO chip without a parent device completely breaks the
ath9k GPIOs for device tree targets.
As long as boards using the devicetree don't have the gpio-controller
property set for the ath9k node, the unloading of the driver works as
expected.
Register the GPIO chip with the ath9k device as parent only for OF
targets to find a trade-off between the needs of driver developers and
the broken LEDs and buttons seen by users.
Fixes: FS#2098
Signed-off-by: Mathias Kresin <dev@kresin.me>
(cherry picked from commit d35f2a5565)
This pick most of brcmfmac changes backported into the master in commits
5932eb690f ("mac80211: brcmfmac: backport firmware loading cleanup")
3eab6b8275 ("mac80211: brcmfmac: backport NVRAM loading improvements")
529c95cc15 ("mac80211: brcmfmac: fix use-after-free & possible NULL pointer dereference")
It's more than would be normally backported into a stable branch but it
seems required. Firmware loading cleanups are needed to allow fix memory
bugs in a reliable way. Memory fixes are really important to avoid
corrupting memory and risking a NULL pointer dereference.
Hopefully this stuff has received enough testing in the master.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
First one is a fix for reporting channels to the user space. Important
for users as they could try setting invalid channel and fail to start an
interface.
Later is a support for newer FullMAC chipset firmwares.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Currently, dynack was only tested upstream using AP/STA mode.
Testing it on IBSS, showed that late-ack detection was broken.
This is caused due to dynack using Association Request/Response
frames for late-ack detection, which IBSS does not use.
Also allowing Authentication frames here solves this.
A second issue also got fixed, which was also seen AP/STA mode:
When a station was added, the estimated value would be exponentially averaged
using 0 as a starting point.
This means that on larger distances, the ack timeout was still not high
enough before synchronizing would run out of late-ack's for estimation.
Fix this by using the initial estimated value as a baseline
and only start averaging in the following estimation rounds.
Test setup:
- 2x identical devices: RB912UAG-5HPnD + 19dB sector
- IBSS
- 2x2 802.11an (ar9340), HT20, long GI
- RSSI's -70 / -71
- Real distance: 23910 meter
Results (60s iperf runs):
Fixed coverage class 54 (up to 24300m):
* 21.5 Mbits/sec
Dynack:
* 28.9 Mbits/sec
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
Software crypto wasn't working for management frames because the flag
indicating management frame crypto was missing
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Registering a GPIO chip with the ath9k device as parent prevents unload,
because the gpiochip core increases the module use count.
Unfortunately, the only way to avoid this at the moment seems to be to
register the GPIO chip without a parent device
Signed-off-by: Felix Fietkau <nbd@nbd.name>
This allows driver to support features that can't be dynamically
discovered.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit fecbd91c7c)
Monitor mode isn't supported yet with brcmfmac, it's just an early work.
This also prepares brcmfmac to work stable with new firmwares which use
updated struct for passing STA info.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit c0608c6a27)
The most important is probably regression fix in handling platform
NVRAM. That bug stopped hardware from being properly calibrated breaking
e.g. 5 GHz for Netgear R8000.
Other than that it triggers memory dumps when experiencing firmware
problems which is important for debugging purposes.
Fixes: 2811c97803 ("mac80211: backport brcmfmac firmware & clm_blob loading rework")
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit b26214adb5)
Most of the implementations behind cfg80211_get_station will not initialize
sinfo to zero before manipulating it. For example, the member "filled",
which indicates the filled in parts of this struct, is often only modified
by enabling certain bits in the bitfield while keeping the remaining bits
in their original state. A caller without a preinitialized sinfo.filled can
then no longer decide which parts of sinfo were filled in by
cfg80211_get_station (or actually the underlying implementations).
cfg80211_get_station must therefore take care that sinfo is initialized to
zero. Otherwise, the caller may tries to read information which was not
filled in and which must therefore also be considered uninitialized. In
batadv_v_elp_get_throughput's case, an invalid "random" expected throughput
may be stored for this neighbor and thus the B.A.T.M.A.N V algorithm may
switch to non-optimal neighbors for certain destinations.
Signed-off-by: Sven Eckelmann <sven.eckelmann@openmesh.com>
(cherry picked from commit 87493dac11)
After a very enlightening but unfortunately far too short exchange with Jes
we mutually agreed to drop the patches. They are unfortunately not ready
yet.
Acked-by: Rafał Miłecki <rafal@milecki.pl>
Signed-off-by: John Crispin <john@phrozen.org>
(cherry picked from commit 66c5696cdf)
It was described by Arend as:
> This series is intended for 4.17 and includes following:
>
> * rework bus layer attach code.
> * remove duplicate variable declaration.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit 0da9303e5b)
When the 160mhz width is selected the ath10k firmware crash. This fix this problem.
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
(cherry picked from commit 134e832814)
Import a revert-commit from Stanislaw Gruszka which significantly
improves WiFi performance on rt2x00 based hardware.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
(cherry-picked from commit f4a639a3d7)
This patch was revised upstream before being merged, and OpenWrt's copy
was never updated to reflect the revision.
Signed-off-by: Thomas Hebb <tommyhebb@gmail.com>
[refreshed patches]
Signed-off-by: Mathias Kresin <dev@kresin.me>
Merge upstream patch from Sathishkumar Muruganandam
<murugana@codeaurora.org> for OpenWrt's backports package:
https://lkml.kernel.org/r/<1522049641-19521-1-git-send-email-murugana@codeaurora.org>
Commit-ID: 606204bb863fa3b0bb54929d79b4dc46338f9180
* FW has Smart Logging feature enabled by default for detecting failures
* and processing FATAL_CONDITION_EVENTID (36925 - 0x903D) back to host.
*
* Since ath10k doesn't implement the Smart Logging and FATAL CONDITION
* EVENT processing yet, suppressing the unknown event ID warning by moving
* this under ATH10K_DBG_WMI.
*
* Simulated the same issue by having associated STA powered off when
* ping flood was running from AP backbone. This triggerd STA KICKOUT
* in AP followed by FATAL CONDITION event 36925.
*
* Issue was reproduced and verified in below DUT
* ------------------------------------------------
* AP mode of OpenWRT QCA9984 running 6.0.8 with FW ver 10.4-3.5.3-00053
*
* Signed-off-by: Sathishkumar Muruganandam <murugana@codeaurora.org>
Signed-off-by: Stefan Lippers-Hollmann <s.l-h@gmx.de>
When CONFIG_PM was not set rsi_sdio_reinit_device() was not compiled
into the driver but referenced.
This is a backport form the mainline Linux kernel.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
RedPine RS9113 wireless module requires rsi91x driver to be built
and linux-firmware/rsi/rs9113_wlan_qspi.rps to be installed.
Also we add patch for successful compilation of rsi91x driver.
Signed-off-by: Evgeniy Didin <Evgeniy.Didin@synopsys.com>
Cc: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Hauke Mehrtens <hauke@hauke-m.de>
Cc: John Crispin <john@phrozen.org>
This patch adds the patch that was posted to ath10k-devel ML:
<https://patchwork.kernel.org/patch/10233491/>
|From: Thomas Hebb <tommyhebb@gmail.com>
|Subject: [PATCH] ath10k: search all IEs for variant before falling back
|Date: Wed, 21 Feb 2018 11:43:39 -0500
|[...]
|This patch fixes the issue by first searching the entire file for the ID
|with variant, and searching for the fallback ID only if that search
|fails. It also includes some code cleanup in the area, as
|ath10k_core_fetch_board_data_api_n() no longer does its own string
|mangling to remove the variant from an ID, instead leaving that job to a
|new flag passed to ath10k_core_create_board_name().
|
|I've tested this patch on a QCA4019 and verified that the driver behaves
|correctly for 1) both fallback and variant BDFs present, 2) only fallback
|BDF present, and 3) no matching BDFs present.
|
|Fixes: 1657b8f84ed9 ("ath10k: search SMBIOS for OEM board file extension")
|Signed-off-by: Thomas Hebb <tommyhebb@gmail.com>
Note: 937-ath10k-calibration-variant.patch has been reassigned a new 081
number, as it now ships with upstream.... But also because this patch
requires the change in ath10k_core_create_board_name().
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
